Microfluidic Biofabrication of 3D Multicellular Spheroids by Modulation of Non-geometrical Parameters

Abstract

Three-dimensional (3D) cell spheroids are being increasingly applied in many research fields due to their enhanced biological functions as compared to conventional two-dimensional (2D) cultures. 3D cell spheroids can replicate tissue functions, which enables their use both as in vitro models and as building blocks in tissue biofabrication approaches. In this study, we developed a perfusable microfluidic platform suitable for robust and reproducible 3D cell spheroid formation and tissue maturation. The geometry of the device was optimized through computational fluid dynamic (CFD) simulations to improve cell trapping. Experimental data were used in turn to generate a model able to predict the number of trapped cells as a function of cell concentration, flow rate, and seeding time. We demonstrated that tuning non-geometrical parameters it is possible to control the size and shape of 3D cell spheroids generated using articular chondrocytes (ACs) as cellular model. After seeding, cells were cultured under perfusion at different flow rates (20, 100, and 500 μl/min), which induced the formation of conical and spherical spheroids. Wall shear stress values on cell spheroids, computed by CFD simulations, increased accordingly to the flow rate while remaining under the chondroprotective threshold in all configurations. The effect of flow rate on cell number, metabolic activity, and tissue-specific matrix deposition was evaluated and correlated with fluid velocity and shear stress distribution. The obtained results demonstrated that our device represents a helpful tool to generate stable 3D cell spheroids which can find application both to develop advanced in vitro models for the study of physio-pathological tissue maturation mechanisms and to obtain building blocks for the biofabrication of macrotissues.

Keywords: 3D culture; fluid dynamic; microfluidic; pellet culture; spheroid.

FIGURE 1

Optimization of the microfluidic chip design. (A) Geometric discretization of the concave chamber used in the computational model. Section of the mesh corresponding to the longitudinal mirror plane, and zoom of a mesh detail. (B) Colorimetric map of velocity magnitude relative to the longitudinal mirror plane and comparison between mean velocity values computed within the simplified and the actual chamber geometry at different depths (flow rate 2.5 μl/min). (C) Comparison between two different configurations of the chip chamber (i.e., without or with a 0.2 mm-high groove). Meshing scheme and colorimetric maps of velocity magnitude relative to cross-sections of the concave chamber at different depths (flow rate 2.5 μl/min).

FIGURE 2

Microfluidic chip fabrication. (A) Schematic of the fabrication process. The top layer mold was obtained by transferring the channel layout onto a silicon wafer by photolithography. PDMS was casted and cured (I). The PDMS layer was detached (II) and inlet/outlet ports were punched (III). The bottom layer was obtained through multiple fabrication steps. The chamber and the groove were machined on a PMMA substrate with a CNC milling machine. PDMS was poured on the PMMA mold and cured (IV). Subsequently, PDMS was detached and used as mold for PDMS casting (V). The PDMS layer was then detached (VI) and bonded to the top layer. (B) 3D rendering and lateral view of the assembled microfluidic device and top view of the concave chamber (scale bars: 1 mm).

FIGURE 3

Cell trapping analysis. (A) Schematic showing the experimental set-up. The tested experimental conditions are represented as a 3D matrix. (B) Representative side views showing the progressive cell trapping in the chamber for different cell concentrations and flow rates. (C–H) Bubble graphs showing the average number of cells trapped in the chamber in correspondence of specific combinations of cell concentrations, flow rates, and seeding times. The bubble size is proportional to the number of trapped cells indicated. (C,D) Number of trapped cells obtained using different flow rates and seeding times starting from 10M or 20M cells/ml. (E,F) Number of trapped cells obtained for different cell concentrations and seeding times applying 2.5 or 5 μl/min. (G,H) Number of trapped cells obtained using parameter combinations theoretically leading to the same outcome. All the experimental conditions were tested independently at least six times (*p < 0.05; **p < 0.01).

FIGURE 4

Cell trapping model. (A) All the experimental conditions included in the model are represented as 2D matrices. (B–E) Interaction plots describing the relation between cell concentration and flow rate or seeding time. Lines describe the variation of the number of trapped cells in correspondence of the minimum and maximum values of the variable indicated in the right y-axis for increasing values of the variable indicating in the x-axis. (F) Schematic of the working principle of the predictive model. (G) Comparison between the numbers of trapped cells obtained experimentally and predicted by the model for the following parameters: 10M cells/ml, 2.5 μl/min, 60 min. (H) Correlation plot describing the correlation between the number of trapped cells experimentally measured and the number predicted by the model.

FIGURE 5

3D cell spheroid formation under different perfusion conditions. (A) Schematic showing the experimental set-up and the tested experimental conditions. (B) Side view of the concave chamber showing aggregate formation over time. (C) Representative pictures showing the process of spheroid formation and colorimetric maps of shear stresses applied on the concave chamber filled with cells under different perfusion conditions (scale bars: 1 mm).

FIGURE 6

Comparison of 3D aggregates generated under different perfusion conditions. (A–C) Quantification of cell number, normalized metabolic activity, and GAG content in 3D articular chondrocyte spheroids after 14 days of culture (*p < 0.05). (D) Hematoxylin/eosin (upper images) and Alcian Blue (lower images) staining of spheroids cultured in perfusion for 14 days under different flow regimen and respective velocity fields and shear stress colorimetric maps (scale bars: image 100 μm; inset 500 μm).